Specific Aims: This study's long-term goal is to improve colorectal cancer care for the elderly who have a high incidence of this highly treatable and potentially curable disease. In 1990, a National Institutes of Health Consensus Panel recommended adjuvant therapies for colorectal cancer patients with stage III colon and stage II and III rectal cancer. The degree to which the elderly, especially traditionally vulnerable groups, are receiving these recommended treatments is unclear. This study will (1) compare the receipt of recommended colorectal cancer treatments between African-American and Hispanic versus white elderly, and rural versus urban elderly; (2) compare diffusion of these treatments from 1992-1996 between our study groups; (3) identify the physician, patient, hospital, and environmental factors that predict differences found in treatment or diffusion; (4) measure treatment cost differences between our study groups, and the cost implications to the Medicare program of providing recommended treatments to all beneficiaries with colorectal cancer; and (5) evaluate several methodologies for cancer research using administrative databases, including different measures of comorbidity and costs. Research Design and Methods: This study will use the linked SEER-Medicare claims database in a retrospective cohort design to examine differences in receipt, diffusion, and cost of recommended colorectal cancer treatments between more and less vulnerable elderly populations. Using stage III colon and stage II and III rectal cancer cases identified in SEER between 1992 and 1996, we will identify different patterns of treatment and alternative measures of the costs for colorectal cancer care with Medicare claims data. Annual study cohorts will be tracked in a comprehensive database using 1991-1998 linked SEER-Medicare, AMA Masterfile, AHA, and Area Resource File data. Physician, hospital, and patient characteristics (including cancer stage and comorbidity) will be linked to the location, size, and socioeconomic profile of places of residence, travel distances to different treatment sites and cancer specialists, service utilization, and allowed charges. In the second phase of the study, we will use multivariate analysis to identify the degree to which physician, patient, hospital and environmental factors predict systematic differences in treatment patterns between our study populations. The cost analysis will include an estimate of the resources required to bring all patients up to the recommended standard of care. Alternate methods of defining comorbidity and costs will be used to evaluate their utility in cancer research.